Gallup To Explain How It Predicted 2012 Presidential Election So Badly
Gallup plans to release a detailed analysis over how its polls during the 2012 election turned out to be so disastrously wrong.
The noted polling company ran against the pack for most of the time before the election, showing Mitt Romney in a consistent lead over Barack Obama even as others showed Obama with a small but widening lead.
Gallup’s errors weren’t just a short-term blip. The company’s daily tracking poll showed Romney with a 1- to 7-point lead down the stretch, after the GOP candidate had a strong showing in the first presidential debate.
Just 10 days before the election the Gallup’s daily tracking poll showed a 4-point lead for Romney, even though Obama would go on to win the election 51 percent to 47 percent.
Shortly after the election concluded, Gallup editor-in-chief Frank Newport pledged to conduct a thorough review of the polling company’s telephone survey methodology. He has pledged to make those findings public in early June.
“We take it seriously” when polls are off, Newport said. “We’ve been doing presidential polling since 1936 which is what put George Gallup on the map …The results [in 2012] certainly were not what we wanted them to be from Gallup’s perspective.”
Gallup’s errors were compounded by the rise of polling analyst Nate Silver, whose blog FiveThirtyEight aggregated polls to make more wide-ranging predictions.
Taking data from hundreds of polls, Silver used a mathematical formula to show Obama holding a small but significant lead throughout the election, even after Romney’s strong debate showing.
Newport has hinted that Gallup’s future could include polling informed by other sources, including “non-probability data, administrative data, social media data.”
“I think it is an exciting time,” Newport said. “Rather than doing things the same way in any business, it’s always exciting when technology in particular opens up all these other opportunities.”
Gallup’s analysis about its 2012 failures will be released June 4.
